Obstet Gynecol Clin N Am 31 (2004) xv–xvi Preface Endocrinology of Pregnancy Aydin Arici, MD Joshua A. Copel, MD Guest Editors Recent expansion of biomedical knowledge on the interactions between the fetus, placenta, and the mother have transformed our view of pregnancy in general. Recent basic and clinical investigations have improved significantly our understanding on how hormones affect the pregnancy, and on how pregnancy affects the fetal and maternal hormones. Because pregnancy may be seen as the ultimate hormonally mediated event, the topic of endocrinology of pregnancy is particularly relevant. The expansion of knowledge that has occurred during the last two decades has pushed medicine toward subspecialization. On the other hand, a general obstetrician-gynecologist is continuously facing challenges to resolve endocri- nologic problems during pregnancy. It is well known that physiologic changes of pregnancy may mask clinical findings and laboratory results of endocri- nologic problems. The endocrinology of pregnancy has become one of the areas that straddles multiple specialties; the authorship of this issue reflects this. The aim of this issue is to present a concise review of latest knowledge on the endocrinology of pregnancy to the reader. One needs to gather experts in perinatology, reproduc- tive endocrinology, medical endocrinology, and neonatology to address topics that are quite broad in scope. A diverse group of internationally recognized ex- perts have come together to discuss the cutting edge knowledge in their 0889-8545/04/$ – see front matter D 2004 Elsevier Inc. All rights reserved. doi:10.1016/j.ogc.2004.09.004 obgyn.theclinics.com xvi A. Arici, J.A. Copel / Obstet Gynecol Clin N Am 31 (2004) xv–xvi respective specialties. We are grateful to all of the authors, all of whom took the time to contribute to this issue despite their other responsibilities. Finally, we greatly appreciate the support of Carin Davis and the staff at Elsevier for their outstanding editorial competence. We hope that this issue will serve women with their babies as well as the physicians who care for them. Aydin Arici, MD Division of Reproductive Endocrinology and Infertility Department of Obstetrics, Gynecology, and Reproductive Sciences Yale University School of Medicine 333 Cedar Street P.O. Box 208063 New Haven, CT 06520-8063, USA E-mail address: Obstet Gynecol Clin N Am 31 (2004) 727–744 Luteal phase defect: myth or reality a b, Orhan Bukulmez, MD , Aydin Arici, MD * a Division of Reproductive Endocrinology and Infertility, Department of Obstetrics and Gynecology, The University of Texas Southwestern Medical Center at Dallas, 5323 Harry Hines Boulevard, Dallas, TX 75390-9032, USA b Division of Reproductive Endocrinology and Infertility, Department of Obstetrics and Gynecology and Reproductive Sciences, Yale University School of Medicine, 333 Cedar Street, New Haven, CT 06510, USA Luteal phase defect (LPD) was described by Jones in 1949 [1]; it is char- acterized by failure to develop fully mature secretory endometrium. This entity is defined as a defect of the corpus luteum to secrete progesterone in high enough amounts or for too short a duration. This results in an inadequate or out-of-phase transformation of the endometrium which precludes embryo implantation. Therefore, LPD is believed to be a cause of infertility and spontaneous mis- carriage. Abnormalities of the luteal phase have been found in 3% to 10% of the female population that has primary or secondary infertility and occurs in up to 35% of those who have recurrent abortion [2]. As a clinical entity, however, LPD is poorly characterized. LPD may be identified in many women who have proven fertility. There is no definite con- sensus in the diagnosis of the condition. Some investigators emphasize the importance of endometrial histology in diagnosis and claim that the actual serum progesterone levels have no value as long as the endometrium is in-phase. Other investigators however, believe that only progesterone levels that are greater than a certain threshold can assure the optimal preparation of endometrium for implantation. LPD also has been believed to be one of the stages of ovulatory disturbance that starts with anovulation and continues as oligo-ovulation, LPD, and normal ovulation [3]. This article reviews the controversies that sur- round LPD. * Corresponding author. E-mail address: 728 O. Bukulmez, A. Arici / Obstet Gynecol Clin N Am 31 (2004) 727–744 Issues in etiopathogenesis The proposed mechanisms of LPD include decreased levels of follicle- stimulating hormone (FSH) in follicular phase, abnormal luteinizing hormone (LH) pulsatility, decreased levels of LH and FSH during the ovulatory surge, decreased response of endometrium to progesterone, and elevated prolactin levels [4]. Furthermore, LPD has been linked to several factors (eg, inadequate endo- metrial progesterone receptors and endometritis) and drugs (eg, clomiphene citrate, gonadotropin releasing hormone (GnRH) agonists and antagonists). Some investigators reported increased LH pulse frequency and abnormal follicular phase LH:FSH ratio [5], whereas others claimed inadequate LH surge [6] as possible etiologic factors for LPD. These findings were not confirmed in other studies [7,8]. Reported follicular phase FSH deficiency with decreased preovulatory estradiol levels as a cause for LPD [6] also was not demonstrated by other investigators [8,9]. Approximately one half of all LPDs have been attributed to the improper function of the GnRH pulse generator in the hypothalamus [10]. Following ovulation, the increased serum progesterone levels oversuppress the GnRH pulse generator which results in too few LH pulses, and therefore, improper luteal function. Hyperprolactinemia has also been implicated in LPD by interfering with GnRH secretion. Latent hyperprolactinemia by interfering with GnRH also has been associated with LPD [10]. In a primate model, 12-day physical and psychologic stress challenge induced LPD which was marked by the decrease in area under the curve for luteal phase serum progesterone levels. The reduction in overall luteal phase progesterone secretion was not associated with a shorter luteal phase which indicated that premature luteolysis did not occur. This reduction however was attributed to the observed decrease in luteal LH levels, which was ultimately related to the stress- induced dysfunction of the hypothalamic-pituitary-adrenal axis [11]. Mild hyper- prolactinemia and exaggerated prolactin release in response to stress also has been associated with LPD or short luteal phase [10,12]. Experimental interference with the profile of gonadotropic stimulation during the follicular phase of the cycle by either using a GnRH agonist [13] or adminis- tering a crude follicular fluid preparation [14] reduced the progesterone secretion during the luteal phase. Other investigators demonstrated a decrease in immuno- reactive FSH levels during the follicular phase in patients with LPD diagnosed by endometrial histology [15]. After the normal folliculogenesis, progesterone secretion can be decreased by interference with gonadotropic support by GnRH antagonist administration during the midluteal phase [16,17]. Abnormal LH pulse frequency has been linked to LPD [18]. LPD also has been associated with decreased inhibin levels in the follicular phase and a subnormal midcycle LH surge [4]. In the corpus luteum, the most abundant cell types are endothelial cells and the pericytes. Resident cells that stem from white blood cell line and fibroblasts also are present [19]. Only a minority of cells are the steroidogenic cells which are of O. Bukulmez, A. Arici / Obstet Gynecol Clin N Am 31 (2004) 727–744 729 two types [20,21]. The large luteal cells originate from the follicular granulosa cells. These cells are not responsive to LH but produce several autocrine and paracrine peptides and eicosanoids. They also produce progesterone and estradiol, in turn, guaranteeing the basal production of these two hormones. The second cell type is the small luteal cells that are derived from the follicular theca cells. These cells acquire LH receptivity and respond to LH pulses with increased estradiol and progesterone secretion. In some patients, LPD is believed to be related to the failure of small luteal cells to respond to LH [10]. An ovarian cause for LPD—in the form of accelerated luteolysis—was suggested as one of the mechanisms [9]. The reasons for early luteal regression were linked to white blood cells and cytokines that are involved actively in the corpus luteum [22,23]. It is clear that any disturbance of ovulatory function may produce LPD in the research setting. The question remains whether each or some of these factors in a given individual is persistent enough to cause ‘‘chronic’’ LPD that leads to infertility or recurrent miscarriage. Diagnosis The optimal means of diagnosing LPD is controversial. It is defined his- torically as a lag of more than 2 days in the histologic development of endo- metrium compared with the day of the cycle. This lag should occur in more than one cycle. Several indicators and laboratory findings have been proposed for the diagnosis of LPD. These include shortened luteal phase in basal body tempera- ture (BBT) charts, decreased luteal phase serum progesterone levels, and dis- crepancies in endometrial histologic findings. Basal body temperature chart BBT measurements were claimed to be useful in the diagnosis of short luteal phase; however, controversy exists regarding the appropriate criteria to use [9]. Progesterone increases the set-point of the hypothalamic thermoregulatory center. A serum progesterone level that is greater than 2.5 ng/mL may increase the BBT up to 18F; this forms the basis of the BBT chart. Traditionally, a biphasic BBT chart with sustained increased temperature for 12 to 15 days is considered to be normal. Determining the length of the luteal phase was proposed to be the simplest approach for the evaluation of luteal function, although its predictive values have been questioned [24]. It was reported that 5.2% of women who have normal ovulatory cycles have luteal phases that are shorter than 9 days [7]. Such luteal phases were observed commonly in women who were younger than 24 and older than 45 years of age. When the temperature elevation is maintained for less than 11 days, the quality of ovulation and the resulting corpus luteum has been considered to be inadequate [7]. In 95 patients who had unexplained infertility, however, there were no differences in the length of the luteal phase when compared with 92 control 730 O. Bukulmez, A. Arici / Obstet Gynecol Clin N Am 31 (2004) 727–744 women who had normal ovulatory cycles [24]. The occurrence of luteal phase duration of up to 11 days were 9% and 8% in women who had unexplained infertility and in controls, respectively [24]. In 30 regularly menstruating women, different BBT patterns and luteal phase lengths were found in 36% and 67% of the observed consecutive cycles, respec- tively [25]. In addition, estrogen and progesterone levels and endometrial dating showed substantial variability in the consecutive cycles of each patient. This in- dicates that the conditions of the luteal phase are not the same in every cycle. In studies, neither the rate of increase in the postovulatory temperature nor the magnitude of temperature elevation correlated with endometrial histology. The overall correlation of BBT charts with endometrial histology was as low as 25% [26]. BBT charts are not reliable enough to be considered as the diagnostic tool for LPD. Endometrial histology The original description of LPD in 1949 incorporated BBT charts, urinary pregnanediol levels, and endometrial biopsy as diagnostic tests [1]. The classic approach to diagnose LPD uses the histologic dating method of Noyes et al [27,28] in endometrial biopsy specimens. This original criterion was described in relation to BBT charts. Reproducibility to within 2 days of BBT charts was obtained in more than 80% of the 8000 biopsy specimens that were studied. The diagnosis is made histologically when endometrial maturation lags 2 or more days behind the expected day of ovulation and the subsequent onset of menses [29,30]. With this technique, the prevalence of LPD in an infertile population has ranged from 3.5% to 38.9% [30–32]. The optimal time for performing an endometrial biopsy has not been determined. In an earlier study, nearly one half of the abnormal endometrial biopsies that were performed during the midluteal phase had reverted to normal when repeated in the late luteal phase [33]. Some investigators recommended late luteal biopsy 11 to 12 days after positive urinary LH testing, although the endo- metrial histology may be increasingly variable as menstruation approaches [3]. When retrospective and prospective dating methods for the diagnosis of LPD were compared, the retrospective method (determination of LH peak by daily assay) identified 42% of biopsy specimens as out-of-phase, whereas the pro- spective method (calculation based on the onset of next menstrual period) iden- tified only 10% as out-of-phase [34]. The results of repeat endometrial biopsies vary during each cycle in the same patient by 15 to 30% [35]. Therefore, two out- of-phase endometrial biopsies from two cycles have been recommended for the diagnosis of LPD. There also has been a disagreement over whether to use a 2-day lag or a greater than 2-day lag to diagnose LPD. Five regularly menstruating women of proven fertility underwent a total of 39 endometrial biopsies [36]. Using a 2-day or greater lag in endometrial maturity to define LPD, the incidence of single and sequential out-of-phase endometrial biopsies was 51.4% and 26.7%, respectively. O. Bukulmez, A. Arici / Obstet Gynecol Clin N Am 31 (2004) 727–744 731 Using a 3-day or greater lag to define a LPD, the incidence of single and sequential out-of-phase endometrial biopsies was 31.4% and 6.6%, respectively. Furthermore, these incidences in normal, fertile women were close to the rates observed in infertile populations [36]. There is significant inter- and intraobserver variability in the results of histologic dating. The duplicate endometrial biopsies from 25 women were dated by five evaluators on two separate occasions [37]. Inconsistencies between the evaluators accounted for 65% of the observed variability, whereas 27% was due to inconsistencies in duplicate readings by the same evaluator [37]. The sig- nificant inter- and intraobserver variability in the results of histologic dating, the issue of cycle-to-cycle variation of biopsy results, the debates in the proper timing of the biopsy, the disagreements over the diagnostic criteria of days of lag in the specimen, and the similar biopsy findings in fertile and infertile women compromise the dependability of endometrial histology in the diagnosis of LPD. Progesterone levels The serum progesterone levels are subject to large fluctuations as a result of pulsatile hormone release [38]. On the basis of a single progesterone determination during the midluteal phase, a false LPD may be diagnosed approximately 15% of the time [10]. Some investigators suggest that because the decreased progesterone levels are seen regularly before the occurrence of an LH pulse, it is more appropriate to draw two or three blood samples within a 3-hour period to decrease the probability of a falsely diagnosed LPD down to 2% to 0.5% [10]. In 457 patients who had regular menstrual cycles and normal ovulation as confirmed by transvaginal ultrasound, the distribution of midluteal phase serum progesterone levels were bimodal with two peaks at approximately 7 ng/mL and 11 ng/mL. The arbitrary cut-off for a normal progesterone level was set at greater than 8ng/mL. Life table analysis of the data showed that the patients who had decreased midluteal progesterone levels had decreased spontaneous fe- cundity [10]. Studies that compared daily luteal serum progesterone levels in women who had unexplained infertility with those who had normal ovulatory or conception cycles reported different cut-off values to define abnormal progesterone levels [39,40]. Some investigators defined abnormal progesterone levels as less than 5 ng/mL for 5 or more days in the luteal phase, whereas other investigators concluded that an abnormal level during the luteal phase was less than 10 ng/mL. The corpus luteum is unresponsive to LH pulses during the early luteal phase. The response to LH develops between Day 4 and Day 6 after ovulation [41]. It has been suggested that if a single determination of progesterone level can be done on one of the days when the corpus luteum becomes responsive to LH, a correct diagnosis of LPD may be more likely [10]. When a midluteal pro- gesterone level of less than 10 ng/mL was considered to be abnormal, the probability of falsely diagnosing LPD was as low as 4% [10]. The same group 732 O. Bukulmez, A. Arici / Obstet Gynecol Clin N Am 31 (2004) 727–744 concluded that LPD may occur in infertile patients at irregular and unknown intervals and may be chronic in only approximately 6% of these women [8]. The use of a single or serial progesterone levels as a diagnostic test has been criticized because of the pulsatile nature of progesterone secretion and the tran- sient decrease in progesterone levels following daily events like food ingestion [42]. Progesterone levels vary up to 10-fold during the 2- to 3-hour pulse interval in the luteal phase [43]. In this respect, multiple daily progesterone measurements with the calculation of integrated progesterone levels during the luteal phase may be more accurate but are not applicable clinically. The sensitivity and specificity of common clinical tests that are used for the diagnosis of LPD were assessed in 58 strictly defined normal women and 34 women who were evaluated for various reasons, including infertility and recurrent abortion [5]. BBT charts, maximum preovulatory follicle sizes, dated endometrial biopsies, and serum progesterone levels (single and multiple) were used in an attempt to predict which patients had decreased integrated progesterone levels during the luteal phase. Luteal integrated progesterone levels—an estimate of total progesterone output over the luteal phase—were determined by summing daily serum progesterone levels starting with the day after the LH surge and ending with the day before the next menstrual period. First, the normal range of integrated progesterone values was determined in a pool of 58 normal volunteers. The investigators calculated an arbitrary cut-off that was inspired from an earlier article that stated the prevalence rate of LPD as 10% [9]. Because 10% of the women in this pool had integrated progesterone values less than 80 ng d days/mL, the cut-off was set as such; however, various cut-off values that were reported in the literature were calculated in a variety of ways in different female populations and were higher than this threshold [12,44,45]. The patient population that was studied, however, had a prevalence rate of LPD of 21% with the cut-off value of less than 80 ng d days/mL [5]. In the study detailed above, unacceptably low sensitivity and/or specificity values were calculated for BBT chart, luteal phase length, and preovulatory follicle diameter for the diagnosis of LPD. Timed endometrial biopsy had mar- ginal sensitivity (29%–57%) and specificity (44%–56%)—whether dated by next menstrual period or midcycle events, which included the day of LH surge or ovulation as determined by ultrasound. The best test for the prediction of decreased integrated progesterone was a single serum progesterone level from the midluteal phase (5 to 9 days after ovulation) that was less than 10 ng/mL (31.8 nmol/L) (sensitivity 86%, specificity 83%) or a sum of three random serum progesterone measurements that was less than 30 ng/mL (95.4 nmol/L) (sensitivity 100%, specificity 80%). The out-of-phase timed endometrial biopsy combined with a single midluteal progesterone level that was less than 10 ng/mL had a sensitivity of 71% and specificity of 93% [5]. In this study, the best dating criterion for endometrial biopsies was next menstrual period rather than the midcycle events. The endometrial biopsy was recommended as a second-line test, especially when LPD needs to be evaluated in a cycle that is treated with ovulation induction or supplemental progesterone [5]. Along with the concerns O. Bukulmez, A. Arici / Obstet Gynecol Clin N Am 31 (2004) 727–744 733 that were described earlier, this study was criticized for using daily measurement of plasma progesterone as the reference test against all other tests that should be assessed [46]. The issues raised were that the receptivity of endometrium to progesterone could vary independent of serum progesterone levels and that histologic delay could be present with physiologic progesterone [47] or despite supraphysiologic progesterone levels [48]. Furthermore, the integrated serum progesterone is not a good indicator of endometrial histology [49]. Measuring urinary pregnanediol glucuronide, a metabolite of progesterone, in the first urine voided daily during the luteal phase was recommended to diagnose LPD. This approach may eliminate variability that is due to pulsatile secretion and may be more indicative of the total progesterone production by the corpus luteum [50–52]. Although this approach is an attractive tool in the research setting, its clinical applicability is difficult. In addition, the proportion of pro- gesterone that is converted and excreted as pregnanediol glucuronide varies with age, stage of menstrual cycle, and other factors [3]. Ultrasound It was recommended to monitor ovarian follicle size with pelvic sonography during the cycle to detect LPD. The follicle diameter was monitored throughout the follicular phase until the day of ovulation; this was indicated by an acute decrease in follicle diameter, abrupt increase in free intraperitoneal fluid, or ap- pearance of intrafollicular echoes. A maximum mean preovulatory follicle diameter of less than 17 mm was considered to indicate LPD [53,54]. In a more recent study, however, a maximum preovulatory follicle size of 17 mm or less was unacceptably insensitive in the diagnosis of LPD [5]. There is no minimum follicle size that separates all normal women from those who have LPD. Studies regarding the assessment of the luteal phase by using transvaginal color and pulsed Doppler ultrasound did not show any significant benefit [55,56]. Clinical conditions that are associated with luteal phase defect Recurrent abortion Recurrent abortion is defined as the loss of three or more consecutive pregnancies before the twentieth week of gestation. This condition may be associated with LPD that is marked by retarded endometrial development in the peri-implantation period. The diagnosis of LPD has been based on the histologic study of a timed luteal phase biopsy according to the method of Noyes et al [27]. In studies that examined timed endometrial biopsy specimens in women who had recurrent abortion, the incidence of LPD ranged from 17.4% [57] to 28% [58]. The evalua- tion of late luteal phase endometrial biopsies that were performed on regularly